Lookahead support is there in my regex engine - the node types NT_SAVE_IPOS and NT_JUMPTO_IPOS are there for that (as well as for \& concats). vim also supports it, although the syntax is different from perl (I think it is \@= to make a zero-width match and some other operator to make zero-width non-match).

Not all of the stuff is working as of yet, but I hope to have it fully working by the end of the week. Only implementation details are missing. I designed the engine by following vim’s dcumentation for regular expressions, and with the determination to make an engine powerful enough to handle all of it.

The main advantage java has has over perl is the posessive quantifier ‘+’.
So whereas J+JF would match JJJJF , J++JF would not. I havent found this overly useful.

When I meant full support, I guess I was thinking about lookaheads and lookbehinds. As in “Easter (?!eggs)\w+” matching “Easter Monday” or “Easter afdasdsadfa” but not “Easter eggs”. This can be damn handy but in the context of an editor you can always do a second filtering operation.

CT

Regarding the non-greedy multi operators, you can see on the listing above NT_OPTIONAL_JUMP_PREF which is the opcode emitted when non-greedy operators are compiled - they are actually already working!

It’s not complete, but only partial implementation stuff is missing - the whole compiler and matcher framework is there and supports everything necessary. It took a lot of effort to get the template-based design right.

What do you mean by “full reg-ex support”? I’m implementing full vi/vim reg-ex support, but I’m not aware of what perl/Java may do that vim doesn’t (although I know some of the syntax details differ).

It would be great to hear what things are needed at this stage.

Thanks for the comments!

This makes quick and dirty text file hacking with reg-exps significantly easier as you can filter out the noise without worrying about it overlapping your targets.

CT.